19 analyses of federal healthcare payment data

About this work

The analyses span hospital cost reports, Part B and Part D public use files, outpatient and physician fee schedules, Medicare Advantage risk adjustment and prior authorization disclosures, IRS and Department of Labor filings, and provider ownership files. The methods recur across them: peer-group outlier analysis, case-mix and risk adjustment, multi-year panels, record linkage across federal sources, and sensitivity ranges with documented limitations.

The work was published as The American Healthcare Conundrum, an independent data journalism series by the firm's principal, which also makes policy recommendations. This page lists the analytic work. All code and data lineage live in one public repository: github.com/rexrodeo/american-healthcare-conundrum.

Medicare program integrity, coding and utilization

1. Medicare Advantage risk-adjustment coding intensity

   Estimates the coding-intensity component of Medicare Advantage payments for 2025, with a V24 versus V28 HCC model sensitivity band, a decomposition of health risk assessment diagnoses, and a state-level allocation. Cross-validated against Kronick et al. (2025) and HHS-OIG health risk assessment audits.

   Data: CMS Geographic Variation PUFs (MA and FFS), CMS HCC model files, HHS-OIG audits, MedPAC · Code, results and methodology

2. Inpatient MS-DRG severity migration, 2013 to 2024

   Twelve-year national panel of Medicare inpatient discharges. A fixed-weight Fisher case-mix index shows coded severity rising 15.5 percent against a 7.8 percent rise in HCC-measured risk; a published patient-level coding-intensity share is applied to the operating payment pool, with confirmed False Claims Act settlements netted out.

   Data: CMS Medicare Inpatient (IPPS) PUFs DY2013 to DY2024, Crespin et al. (2024) · Code, results and methodology

3. Per-contract Medicare Advantage prior authorization

   The first public per-contract dataset extracted from CMS-0057-F prior authorization disclosures: 93 contracts and 18.4 million requests, with volume-weighted denial rates, per-contract variance and appeal overturn rates.

   Data: CMS-0057-F plan disclosures, insurer 10-K filings · Code, results and methodology

4. Low-value care in Medicare

   Full analysis of 268,634 rows of CY2023 Medicare data against a 31-service low-value care list, with state-level P90/P10 variation per beneficiary and a difference-in-differences comparison.

   Data: CMS Provider Utilization, Hospital Outpatient and Geographic Variation PUFs CY2023 · Code, results and methodology

5. Site-of-service payment differentials

   Hospital outpatient (OPPS) versus physician office (PFS) payment rates for the same clinic visits and minor procedures, applied to 2023 utilization.

   Data: CMS OPPS Addendum B and PFS RVU files CY2025, Medicare Physician PUF DY2023 · Code, results and methodology

6. Over-the-counter drugs billed through Part D

   Part D spending on drugs that are available over the counter, by product.

   Data: CMS Part D Spending by Drug 2023 · Code, results and methodology

Drug spending and pricing

1. Part B physician-administered drugs

   A 597-drug universe of single-source Part B drugs built from the CY2023 public use file after documented exclusions, with biosimilar handling, 340B netting and scenario ranges on the international price ratio.

   Data: CMS Part B Spending by Drug PUF CY2023, MedPAC, ASPE, RAND · Code, results and methodology

2. Part D brand drugs outside IRA negotiation

   The next 100 single-source brand products after the IRA-negotiated drugs, priced net-to-net against international benchmarks with category rebates netted once.

   Data: CMS Part D Spending by Drug PUF CY2023, RAND RR-A788-3, NHS Drug Tariff · Code, results and methodology

3. International drug price comparison

   Product-level comparison of US Part D prices with peer-nation prices for the same drugs.

   Data: CMS Part D, NHS Drug Tariff, RAND · Code, results and methodology

4. GLP-1 spending trend and coverage projection

   Five-year spending trend and an original ten-year budget projection for Medicare coverage of GLP-1 drugs.

   Data: CMS Part D, OECD, KFF, CBO · Code, results and methodology

5. 340B: per-hospital discount spread versus charity care

   Per-entity panel of 1,986 disproportionate-share hospitals comparing estimated 340B spread captured with charity care at cost.

   Data: CMS HCRIS, IRS Form 990 Schedule H, HRSA program purchase data · Code, results and methodology

6. Pharmacy benefit manager economics

   Synthesis of federal and state audit findings on spread pricing and rebate retention.

   Data: FTC interim reports, Ohio Auditor of State, peer-reviewed literature · Code, results and methodology

Hospital cost reports and market structure

1. Hospital administrative cost variance

   Original cost-report analysis of 4,518 hospitals showing 6.2-fold variance in administrative overhead, set against national and international benchmarks.

   Data: CMS HCRIS, CMS National Health Expenditure Accounts, OECD · Code, results and methodology

2. Hospital supply cost per discharge

   Case-mix-adjusted supply cost per discharge across 5,480 hospitals, compared within same-size peer groups.

   Data: CMS HCRIS FY2023 · Code, results and methodology

3. Nonprofit hospital tax exemption versus charity care

   Panel of 3,005 nonprofit hospitals joining cost reports to per-filer Form 990 Schedule H data parsed from IRS bulk XML (2,103 filers).

   Data: CMS HCRIS FY2023, IRS Form 990 Schedule H · Code, results and methodology

4. Hospital mergers and market concentration

   Panel of 1,155 ownership-change events, narrowed to 530 horizontal hospital mergers in 315 hospital service areas, with HHI-dependent price effects applied per market.

   Data: CMS Provider of Services files 2018 to 2025, CMS HCRIS, Dartmouth crosswalks · Code, results and methodology

5. Commercial versus Medicare hospital prices

   Commercial-to-Medicare price ratios for identical hospital services.

   Data: CMS HCRIS, RAND Hospital Price Transparency Study 5.1 · Code, results and methodology

Physician payment and employer plans

1. Specialist compensation and RVU valuation

   Productivity-normalized comparison of US specialist pay with 18 high-income countries, with workforce-mix and RVU misvaluation components.

   Data: CMS PFS RVU file CY2025, Medicare PUF 2024, BLS OEWS, OECD · Code, results and methodology

2. Employer health plan broker and administrative fees

   First plan-level analysis of post-2021 broker and administrative fee disclosures: 8,180 health and welfare plans covering 23.8 million participants.

   Data: DOL Form 5500 Schedules A and C 2023, KFF Employer Health Benefits Survey · Code, results and methodology